NET PER's, Total Min, NBA Rank after 64 Games

[Jammer]

The topic of NET PER's has been discussed recently in the All-Defense Team thread.
Interestingly, last year's All-Defense selections track closely to NET OPPONENT PER.

Getting a good (positive) overall NET PER is very difficult,
because it means that you are contributing more than you are hurting,
both on offense and defense.

It's very severe, since a lot of teams are crappy, many players have crappy NET PER's.

Anyway, as far as the Celtics go, weep over the two new arrivals.
35% of Total Team Minutes are Required for an NBA rank,
only 204 NBA players are ranked.

NET PER ***** Player ***************** Total Minutes ********* NBA NET PER Rank
+32.5 ********* J. R. Giddens *************** 7 (still want to trade him ??)
+09.8 ********* Kevin Garnett ********** 1705 ************************* 10
+07.8 ********* Ray Allen **************** 2337 ************************* 15
+06.6 ********* Paul Pierce ************** 2414 ************************* 26
+06.2 ********* Rajon Rondo ************ 2090 ************************* 33 (Age 23)
+04.4 ********* Tony Allen **************** 762
+02.4 ********* Eddie House ************ 1099 ************************** 74
+02.2 ********* Leon Powe ************** 1071

-01.2 ********* Bill Walker ***************** 74 (Age 21)
-02.7 ********* Kendrick Perkins ******* 1677 ************************* 159
-05.1 ********* Gabe Pruitt ************** 319 (Age 23)
-07.0 ********* Glen Davis ************** 1164 ************************** 197 (Age 23)
-10.9 ********* Brian Scalabrine ********* 504
-13.8 ********* Mikki Moore ************** 78 (height ain't everything)
-17.0 ********* Steph Marbury ************ 76 (needs to make some shots, and avoid turnovers)

[SonicYouth34]

PER's are garbage and don't tell crap about a player.

[Joekickass2008]

PER's are garbage and don't tell crap about a player.

Agreed!! this kinda makes Giddens look like our best player.....

And Moore has been effective thus far...I have been impressed with his play on the offensive end.

[OBisHalJordan]

Jammer, you're always posting stuff about the per could you post an article or something that actually explains what it is?

[Jammer]

Jammer, you're always posting stuff about the per could you post an article or something that actually explains what it is?

You calculate the following for each player, each game AND
the for the OPPONENT you ACTUALLY DEFEND each game,
and calculate the difference.

That is why we post NET PER.
Again, it's not simply PER,
it's YOUR PER minus YOUR MAN's PER.

PER =
(League Pace/Team Pace) x (15.00/League Average) x (1/Minutes Played)
x [3 Pt. FG
+(Assists x 0.67)
+(FG made x {2-[(team assists/team FG) x 0.588]})
+(FT made x 0.5 x {1 + [1-(team assists/team FG)] + [(team assists/team FG) x 0.67]})
-(VOP x turnovers)
-(Missed FG x VOP x league DRB%)
-(Missed FT x VOP x 0.44 x [0.44+(0.56 x league DRB%)]}
+(Defensive rebounds x VOP x (1 - league DRB%)]
+[Offensive rebounds x VOP x league DRB%)
+ (steals x VOP)
+ (blocks x VOP x league DRB%)
-{fouls x (lg. FT makes per foul - (lg. FT attempts per foul x 0.44 x VOP)]}]

where VOP = league value of possession
meaning average points per possession before turnovers and fouls

and

League Pace =
{[(Team FT Attempts + Opponent FT Attempts) x 0.44]
+ Team FG Attempts + Opponent FG Attempts
+Team Turnovers + Opponent Turnovers
- offensive Rebounds - Opponent Offensive Rebounds) x 48

All Divided by Team Minutes Times 2

Although I detested PER for years,
after carefully reviewing the numbers, data and results for about 2 years,
while basically understanding the formula used and why it is what is,
well, I came to become a believer in NET PER.

I still have a lot of problems with straight PER,
and for NET PER to be more meaningful, you need significant minutes.
Roland Lazenby suggests 35% of total team minutes as a starting point for ranking players.

If you have further questions I suggest reading
John Hollinger's Book and
spending some time over at http://www.82games.com.

[Jammer]

NET PER's brutal in that if you're commiting fouls, turnovers, missing shots, and not scoring;

your NET PER will be lousy.

If, at the same time, the man you're guarding is scoring points,
YOUR NET PER will be downright horrible.

[Jammer]

PER's are garbage and don't tell crap about a player.

Agreed!! this kinda makes Giddens look like our best player.....

And Moore has been effective thus far...I have been impressed with his play on the offensive end.

Moore's been total crap.
Just like he was in Sac town, only worse.

And Giddens, well, 7 minutes of action isn't enough to judge anyone
one way or another.
But it does tell J. R. played great D and contributed in his 7 minutes of NBA action.

I have posted the overall minutes for each player.

[ParticleMan]

as with any statistic, it's only relevant if you have a large enough sample.

the jury's still out on steph and mikki (though they've been bad so far). and giddens... lol.

[Zin5]

Ditto on the large enough sample thing, but I'd also like to criticize PER a bit. OK, it's great as a broad evaluation of a player but is by no means the be all, end all statistic that some hacks like John Hollinger crack it up to be (of course he heralds it as such, he came up with the thing). Its means should be as a preliminary judge of a player's overall ability, like if you wanted just a quick idea of how good that player is. It's not going to tell the whole story.

[3pt %]

PER, including net PER really doesn't evaluate defence very well.

Even Hollinger alludes to this. There are just very few stats kept for the NBA that allow you to gauge a players on ball and off ball D.

That said, PER is one of the most inclusive of systems for calculating the stats we do have, so it has a fair degree of value.

Just not as much as eyeballing the game.

Thanks for the regular updates J.

[Pogue Mahone]

I appreciate the work Jammer puts into providing updates, too. My main gripe, though, is not with the premise of "what you produce in relation to what your man produces" but rather it is that PER and it's derivatives don't necessarily jive with reality.

It is because PER finds possession-users, regardless of quality, to be very tasty. So long as they are a shot-a-holic and provide even the most modest off-the-ball skills, they will do well in PER. A 16.0 PER on a 15.0 Usg-R is a hell of a lot more valuable to the team than an 18.0 PER on 22.0 Usg-R. Those 7.0 extra possessions per 40 minutes don't just evaporate into the ether; they are redistributed amongst teammates.

If you don't know the Usg-R of a player, PER doesn't really tell much, imo. Or maybe better put, Usg-R provides context for PER. It is a tricky thing, though, because the efficiency vs opportunity conundrum makes it so.

There are two inevitabilities in basketball. To produce requires 1. minutes and 2. opportunity. Simply being on the floor doesn't mean you are getting opportunity, either, because there are a myriad of factors which can affect the synergy of the players on the floor. If you don't see the floor, you will not produce.

For instance, if I am a high-usage wing, who tends to freeze out teammates on the offensive end, break off plays and looks to simply attack via pull-ups off the dribble, I, in all probability, would be called Ricky Dav.... err, I would do well in PER. Now, there is value in providing that skill-set, don't get me wrong but it is not necessarily the best use of possessions and, therefor, not really indicative of quality of play.

If, on the other hand, I am a ball-swinging (damn right I am, ladies), floor-spacing, pick-setting, backdoor cutter-finding, low-usage, redheaded three point shooter, I will likely be ranked one of the worst players in the NBA by PER.

Personally, I would argue the second player provides more to the team because those extra possessions are, more oft than naught, good shots because the low-usage player is working with teammates to find the best shots within the offense. Secondly, that shooter requires opportunity to be created by teammates but, within the context of the offense, is able to find scoring opportunities for teammates. If everyone is doing their job within the roles of the offense, there will be shots available for the floor spreader.

Sure, the counter-argument could be that PER values shot creation, which is fine, but that doesn't really hold true. The reason for that is that shots have to be put up. It doesn't matter if those shots were created by a teammate or by the player himself.

If I am a post player, who doesn't venture very far from the post and require an entry passes before I am unable to initiate a move, well, I am not going to put up shots if the perimeter players on my team are unable to make a simply entry pass consistently. I am not going to be able to generate offense if my shooting guard dribbles with his head down and ignores me on the pick-and-roll. My performance is not going to reflect well in PER if I am playing most of my minutes with high-usage players because of the lack of offensive opportunity.

Now, if I am that same post player and I am asked to guard the toughest of opposing post scorers AND I am the back line of defense for my team, one that has a shooting guard who tends to be susceptible to giving up dribble penetration, my Opp PER is again reliant on teammates to do their job properly. If my teammates don't rotate after I step to clog the SG's man, it reflects poorly on me. If I blitz the pick-and-roll at the top of the key and the SG doesn't hedge in from the perimeter or the PF isn't cognizant of the picker diving to the front of the tin, my Opp PER will suffer.

It stands to reason that any low-usage player who tends to cover high-usage players on defense, will reflect poorly in Net PER simply because of the inherent issues associated with PER and it's inability to properly measure performance in relation to usage, imo.

[Joekickass2008]

PER's are garbage and don't tell crap about a player.

Agreed!! this kinda makes Giddens look like our best player.....

And Moore has been effective thus far...I have been impressed with his play on the offensive end.

Moore's been total crap.
Just like he was in Sac town, only worse.

And Giddens, well, 7 minutes of action isn't enough to judge anyone
one way or another.
But it does tell J. R. played great D and contributed in his 7 minutes of NBA action.

I have posted the overall minutes for each player.

In my opinion despite what the PERS indicate, I think Moore has been a good pick up. Hit deeps jumpers, which is something I didnt think he could. I think he will prove to be a valuable pick up

Oh and sowwy.... but if he was 'Just like he was in Sac town, only worse.' then he wasnt the same as in Sac was he??? Cause he is worse???

[Jammer]

I appreciate the work Jammer puts into providing updates, too. My main gripe, though, is not with the premise of "what you produce in relation to what your man produces" but rather it is that PER and it's derivatives don't necessarily jive with reality.

It is because PER finds possession-users, regardless of quality, to be very tasty. So long as they are a shot-a-holic and provide even the most modest off-the-ball skills, they will do well in PER. A 16.0 PER on a 15.0 Usg-R is a hell of a lot more valuable to the team than an 18.0 PER on 22.0 Usg-R. Those 7.0 extra possessions per 40 minutes don't just evaporate into the ether; they are redistributed amongst teammates.

If you don't know the Usg-R of a player, PER doesn't really tell much, imo. Or maybe better put, Usg-R provides context for PER. It is a tricky thing, though, because the efficiency vs opportunity conundrum makes it so.

There are two inevitabilities in basketball. To produce requires 1. minutes and 2. opportunity. Simply being on the floor doesn't mean you are getting opportunity, either, because there are a myriad of factors which can affect the synergy of the players on the floor. If you don't see the floor, you will not produce.

For instance, if I am a high-usage wing, who tends to freeze out teammates on the offensive end, break off plays and looks to simply attack via pull-ups off the dribble, I, in all probability, would be called Ricky Dav.... err, I would do well in PER. Now, there is value in providing that skill-set, don't get me wrong but it is not necessarily the best use of possessions and, therefor, not really indicative of quality of play.

If, on the other hand, I am a ball-swinging (damn right I am, ladies), floor-spacing, pick-setting, backdoor cutter-finding, low-usage, redheaded three point shooter, I will likely be ranked one of the worst players in the NBA by PER.

Personally, I would argue the second player provides more to the team because those extra possessions are, more oft than naught, good shots because the low-usage player is working with teammates to find the best shots within the offense. Secondly, that shooter requires opportunity to be created by teammates but, within the context of the offense, is able to find scoring opportunities for teammates. If everyone is doing their job within the roles of the offense, there will be shots available for the floor spreader.

Sure, the counter-argument could be that PER values shot creation, which is fine, but that doesn't really hold true. The reason for that is that shots have to be put up. It doesn't matter if those shots were created by a teammate or by the player himself.

If I am a post player, who doesn't venture very far from the post and require an entry passes before I am unable to initiate a move, well, I am not going to put up shots if the perimeter players on my team are unable to make a simply entry pass consistently. I am not going to be able to generate offense if my shooting guard dribbles with his head down and ignores me on the pick-and-roll. My performance is not going to reflect well in PER if I am playing most of my minutes with high-usage players because of the lack of offensive opportunity.

Now, if I am that same post player and I am asked to guard the toughest of opposing post scorers AND I am the back line of defense for my team, one that has a shooting guard who tends to be susceptible to giving up dribble penetration, my Opp PER is again reliant on teammates to do their job properly. If my teammates don't rotate after I step to clog the SG's man, it reflects poorly on me. If I blitz the pick-and-roll at the top of the key and the SG doesn't hedge in from the perimeter or the PF isn't cognizant of the picker diving to the front of the tin, my Opp PER will suffer.

It stands to reason that any low-usage player who tends to cover high-usage players on defense, will reflect poorly in Net PER simply because of the inherent issues associated with PER and it's inability to properly measure performance in relation to usage, imo.

Some great stuff, Professor Pogue.

PER does value assisted baskets less than unassisted baskets, but unfortunately uses some kind
of average so whether your actual basket was assisted or not it doesn't factor in, if I remember correctly. I think it adjusted for the teams overall # of assists per games. So, guys who create their shot and score without assists (Paul Pierce) are penalized while guys catching and shooting off screens (Ray Allen) aren't (although Ray drives the lane quite a bit, also).

That's a nit.

But I totally agree with your comments on usage, in that for years, Allen Iverson
had a NET PER in the +5.0 to +11.0 range; but used up a ton of possessions
(actually, Kobe was doing the same for years).

This year, Allen Iverson's NET PER is a career low +1.5;
while Kobe, in his new efficient mode, is +14.6

You mentioned Ricky Davis before, his NET PER has always sucked
because he was never a great shooter or defender
(Ricky's NET PER this year with the Clippers is -10.4).

[Jammer]

Giddens ridiculously high NET PER won't remain there, but the guy has only played 7 minutes.

Steph and Mikki's will hopefully improve significantly,
actually, they have to, because they are so bad right now.

[LongTimeFan]

I'm a believer in net PER also. It makes an honest attempt at measuring defense on an equal footing with offense. Since defense has been huge attribute of the various Celtic teams, including this one, I believe it more accurately measures a player's total contribution.

Defense is more important and very poorly measured. Fans care about offense and the glitz.

[AlCelticFan]

I'd be interested to see some kind of PER / usage stat. Or better yet, some kind of PER/usage - opponent's PER/ their usage.

[Red2]

have to say that while I like Moore's enthusiam and passion I was hoping for a bit Moore from him but it's still early.

[Pogue Mahone]

I'd be interested to see some kind of PER / usage stat. Or better yet, some kind of PER/usage - opponent's PER/ their usage.

I am loading a schema right now but if you give me a bit, I will throw something together.

[Pogue Mahone]

aPER: PER normalized to league average usage rate
aOppPER: Opposition PER normalized to league average usage rate
NetaPER: Net adjusted PER

Code: Select all

Player      aPER   aOppPER   NetaPER
Garnett, K.   21.31    15.14    6.16
Giddens, JR   22.79    17.01    5.79
Rondo, R.   20.61    15.53    5.08
Powe, L.   18.79    14.38    4.41
Pierce, P.   17.42    13.85    3.57
Walker, B.   15.76    12.40    3.37
House, E.   16.54    13.96    2.58
Allen, T.   14.73    13.12    1.61
Perkins, K.   17.97    17.24    0.73
Allen,R.   18.42    18.00    0.42
Pruitt, G.   10.25    11.02    (0.77)
Moore, M.   10.48    11.98    (1.50)
Davis, G.   14.49    16.24    (1.75)
Scalabrine, B.   13.50    16.36    (2.86)
O'Bryant, P.   9.53    16.70    (7.16)
Marbury, S.   (0.66)   15.12    (15.78)


While I think this chart really doesn't show anything but the inherent biases of PER, it does Harrison Bergeron the numbers a bit to get a better idea of just how much opportunity (usage rate) is obfuscating the play on both sides of the ball.

One thing to keep in mind is that while it is true that low usage players often struggle under the burden of additional possessions, the inverse is also true; high usage players have trouble remaining efficient with less possessions at their disposal. So, while it is probably a bit of a stretch to assume that Kendrick Perkins is going to remain as efficient scoring the ball with more possessions, it is just as likely that Ray Allen or Paul Pierce would lose some efficiency to their respective games by eschewing half a dozen possessions per 40 minutes played.

Even though I believe this to be an improvement over 82games Net PER, at best it is just an approximation/guesstimate of the level of impact on each side of the ball.